Embodied Decision Making on ALFWorld (test)
89.4Task Success RateMAA
Evaluation results
| Method | Links | |
|---|---|---|
| MAABackbone=Qwen3.7-Max2026.06 | 89.4 | |
| MAABackbone=GPT-5.42026.06 | 88.4 | |
| SkillOptBackbone=Qwen3.7-Max2026.06 | 88.1 | |
| SkillOptBackbone=GPT-5.42026.06 | 87.5 | |
| MAABackbone=Qwen3.6-Flash2026.06 | 86.2 | |
| Trace2SkillBackbone=Qwen3.7-Max2026.06 | 85.8 | |
| Trace2SkillBackbone=GPT-5.42026.06 | 85.3 | |
| Reactive UpdateBackbone=Qwen3.7-Max2026.06 | 85.2 | |
| MAABackbone=DeepSeek-V4-Flash2026.06 | 84.9 | |
| Reactive UpdateBackbone=GPT-5.42026.06 | 84.8 | |
| SkillOptBackbone=Qwen3.6-Flash2026.06 | 84.2 | |
| Single-shot (ExpeL-style)Backbone=Qwen3.7-Max2026.06 | 83.5 | |
| SkillOptBackbone=DeepSeek-V4-Flash2026.06 | 83.2 | |
| Single-shot (ExpeL-style)Backbone=GPT-5.42026.06 | 83 | |
| Reactive UpdateBackbone=Qwen3.6-Flash2026.06 | 81.6 | |
| Trace2SkillBackbone=Qwen3.6-Flash2026.06 | 81.2 | |
| Frozen (no memory)Backbone=Qwen3.7-Max2026.06 | 81 | |
| Frozen (no memory)Backbone=GPT-5.42026.06 | 80.5 | |
| Trace2SkillBackbone=DeepSeek-V4-Flash2026.06 | 80.5 | |
| Single-shot (ExpeL-style)Backbone=Qwen3.6-Flash2026.06 | 79.5 | |
| Reactive UpdateBackbone=DeepSeek-V4-Flash2026.06 | 79.4 | |
| Single-shot (ExpeL-style)Backbone=DeepSeek-V4-Flash2026.06 | 77.8 | |
| Frozen (no memory)Backbone=Qwen3.6-Flash2026.06 | 76.8 | |
| Frozen (no memory)Backbone=DeepSeek-V4-Flash2026.06 | 74.3 |